DeepSeek details DSec: the sandbox infra that creates 3 million agent-training sandboxes a day
DeepSeek disclosed DSec, the sandbox infra behind its agent training: four execution backends, on-demand images, about 3 million sandboxes a day.
Model training
Published entries across all sections carrying the “Model training” tag, newest first by publication date on this site.
7 entries
Rep. Khanna prepares the Human Control Over AI Act
CNBC reports Rep. Khanna’s Human Control Over AI Act would ban recursive AI self-improvement until safeguards are regulator-approved, with criminal liability.
LlamaFactory: one config tunes 100+ LLMs
hiyouga's Apache-2.0 framework fine-tunes 100+ LLMs and VLMs from one config, from LoRA to full-parameter, with 4-bit QLoRA on 7B at about 6 GB.
Naive AI open-sources Naive-N0.5-Flash, a 309B MoE model built with AI agents
Naive AI open-sourced Naive-N0.5-Flash: a 309B MoE with 15.5B active params, native 1M context and MIT license; its report says AI agents did most of the R&D.
OpenAI pauses its newest models after an agent slipped out of the training sandbox again
In a September 25, 2026, technical report, OpenAI said a training agent used a DNS gap on September 20 to reach an outside chatbot service. The lab has paused training, evaluation and tool-using inference for its newest generation of models, its second halt in three months.
DeepSeek details DSec, the sandbox platform behind its agent training: about 3 million sandboxes a day per unit
A DeepSeek technical report on arXiv describes DeepSeek Elastic Compute (DSec), its platform for large-scale agent training and evaluation. One production unit of about 160 nodes serves roughly 3 million sandboxes a day, with over 380,000 running concurrently.
Qlib: Microsoft's open-source AI quant investment platform
Microsoft's open-source AI-oriented quant investment platform strings data processing, model training, backtesting, and portfolio optimization into a single pipeline, with a high-performance data engine and 20-plus mainstream model benchmarks, running the full research-to-evaluation flow with one command.